返回首页
AI on Medium··行业媒体

I Planted 6 Attacks in QwenPaw's 18 Tasks — Its Guards Caught 5, and the 6th Is the Scary One

中文摘要

阿里巴巴QwenPaw助手可扫描恶意软件,但在六次攻击测试中,系统漏掉了最危险的一次,暴露了该AI助手严峻的安全漏洞。

English Summary

Alibaba's QwenPaw AI scans for malware, yet testing showed its guards missed one of six attacks. This dangerous bypass reveals critical security gaps in the personal assistant.

原文节选

Alibaba’s AgentScope team shipped a personal AI assistant that scans your skills for malware before they run. I spent a week trying to get… Continue reading on Medium »